Maximizing output power from PLL synthesizer high-frequency open-collector outputs

At the heart of a transceiver is usually a phase-locked-loop (PLL) synthesizer, which delivers a signal to downstream components such as mixers, filters and power amplifiers. At higher frequencies, impedance mismatch and board losses reduce output power delivered by a PLL synthesizer or clocking device potentially resulting in degraded performance such as reduced sensitivity. Thus, a PLL frequency synthesizer needs to deliver high power with output impedance that matches the transmission and the load it is driving in order to provide margin for losses in the printed circuit board (PCB) and downstream components.

This need for higher output power is contradicted by trends in semiconductor process technology toward lower voltages. These trends create a challenge for designers as lower voltages often result in lower output power from the PLL. The open-collector architecture in Figure 1 is a popular solution to this challenge, as it enables most of the integrated circuit (IC) to operate at lower voltages while still delivering higher output power and allowing less heat dissipation in the IC itself.

Figure 1

Typical open-collector output with a 50 Ω pull-up

Typical open-collector output with a 50 Ω pull-up

For the open-collector output, a transistor drives current through a resistor or inductor connected to a higher voltage. This is known as a pull-up component. For the purposes of this article, we can assume that the pull-up component is external to the IC and that it is placed as close to the chip as possible. Since there will be very poor matching between the output of the transistor and the pull-up, the only way to mitigate signal loss due to mismatch is to keep the trace short (less than one-tenth of the wavelength of the output frequency).

50 Ω resistors are a common choice for the pull-up component because they theoretically match the output impedance to the impedance of the transmission line and load. In reality, however, the output impedance is actually less than the desired 50 Ω because transistor output impedance adds in parallel to the 50 Ω pull-up. Furthermore, there will be an undesired DC drop across the pull-up resistor that will start to shut off the output transistor and lead to lower output power than expected. Figure 1 shows a 50 Ω pull-up being used, but in some cases, the resistor may be replaced by an inductor. This article compares the impact of using an inductor as a pull-up component versus a resistor pull-up.

Understanding the impact of the pull-up component on output power ,/p>

To calculate the output power, the pull-up (ZPullUp ) component and the load (50 Ω) connect to AC ground, as shown in Figure 2.

Figure 2

AC analysis of open-collector output

AC analysis of open-collector output

Neglecting all losses and transmission line effects, the voltage at the load would be the current multiplied by the parallel combination of the load and pull-up impedance. From this value, Equation 1 calculates the power delivered to the load as:

It can be inferred from Equation 1 that increasing the impedance of the pull-up component improves the output voltage swing and therefore the output power. However, if a resistor is used as the pull-up component, then increasing the value of this may backfire because it can create a DC voltage drop across this pull-up resistor and start to shut off the transistor. A better approach when the output is operating at a higher frequency is to use an inductor since it enables increased impedance without creating a DC voltage drop. Figure 3 shows the application of Equation 1 to calculate the relative power for pull-up values of 50 Ω, 1 nH and 10 nH.

Figure 3

Theoretical relative output power compared to a 50 Ω pull-up resistor

Theoretical relative output power compared to a 50 Ω pull-up resistor

Figure 3 shows that an inductor pull-up theoretically delivers higher output power relative to the 50 Ω pull-up for frequencies above 4.5 GHz. This advantage of higher output power is theoretically approaches 6 dB provided the frequency or inductor value is sufficiently high. However, there are real-world component properties, such as the self-resonant frequency of the inductor, that result in practical limitations. Figure 4 shows actual measured data from the Texas Instruments LMX2594 connected to the 50 Ωinput of a spectrum analyzer. The 1-nH inductor had a self-resonant frequency of 13.6 GHz and the 10-nH inductor had a self-resonant frequency of 3.8 GHz. At 1 GHz, the 1-nH inductor has lower power due lower impedance, as predicted in Figure 3. At higher frequencies, both the 1-nH and 10-nH inductors outperform the 50 Ωresistor. Past 10 GHz, the 1-nH inductor starts to outperform the 10-nH inductor due to the nonideal properties of the 10-nH inductor above its self-resonant frequency.

Figure 4

Single-ended output power vs. pull-up component type for the TI LMX2594 synthesizer

Single-ended output power vs. pull-up component type for the TI LMX2594 synthesizer

The inductor pull-up offers an improvement in output power above the 50 Ωresistor, but has two drawbacks:

  • Low output power at low frequency
  • Poor impedance matching

For cases where the output frequency is low, a pull-up resistor may be the preferred option due to its higher power and better matching. However, at higher frequencies, a pull-up inductor can offer higher output power. If the output is operating at low frequencies, one additional modification to the pull-up inductor approach is to add a series resistor (as shown in Figure 5) to address concerns of low impedance at low frequencies and behavior at or above the self-resonant frequency. Appropriate values for this resistor would range from 25 to 50 Ω

Figure 5 also adds a resistive pad to improve matching. For instance, if the pull-up inductor has poor matching but improves power by 6 dB, then a 3-dB pad could be use and still provide a 3-dB benefit. In addition, the pad makes the output less sensitive to loading effects, providing output impedance closer to 50 Ω even if the impedance of the transistor’s collector node is not high relative to the pull-up component.


When used at a high frequency, open-collector outputs can help maximize power in a PLL synthesizer, helping overcome industry trends to lower power ICs. However, it is important to be cautious when assuming that a pull-up resistor of 50 Ω is the best choice for matching or output power – especially when the pull-up component is external to the IC. It may be the case that selecting a pull-up inductor along with a resistive pad could offer higher output power and better impedance match.

1 comment on “Maximizing output power from PLL synthesizer high-frequency open-collector outputs

  1. AaronHe
    March 13, 2019

    Hi Dean:

    How about set the pull-up components as an inductor parallelled with a resistor for example 50 ohm. Choose the inductor value big enough at the working frequency to make the parallelled impedance is almost same as the resistor. And this added inductor will remove the DC voltage drop which could guarantee the output signal will have full voltage swing.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.